Range predicting where a single new study's true effect would likely fall.
Unlike a confidence interval, which estimates the precision of the pooled average effect, a prediction interval estimates the range within which a new, similar study's result would plausibly fall, accounting for between-study heterogeneity. A prediction interval crossing zero signals that some future studies or contexts could plausibly show no effect, even when the pooled estimate looks favourable.
This guide was auto-drafted and is pending editorial review.